CaratShares Journal · Auction watch

A $34M auction just repriced rubies. Or did it?

By Alex Chubuk, founder of CaratShares · September 2026 · 4 min read

On September 17, Sotheby's Hong Kong High Jewelry sale hammered through more than 180 lots for a reported HK$267.5 million — about $34.1 million. The top lot, a Mauboussin necklace set with tumbled rubies and old European-cut diamonds, brought HK$16 million (~$2 million) against a high estimate of HK$11 million: roughly 45% over.

If you sell, insure, appraise or collect coloured stones, here is the uncomfortable question: what did you just learn about the price of rubies?

What the hammer actually says

A 45% beat over estimate sounds like a signal. But over-estimate results conflate at least four things: the estimate itself (a marketing artefact, set to entice), the specific object (a signed Mauboussin piece is jewellery-house provenance, not a loose-stone comparable), the room (two motivated bidders on one afternoon), and the season (Hong Kong autumn sales have been strong for two years running). Which of those moved? The auction result cannot tell you — it is one number carrying four confounded variables.

An auction is a price event. A market is a price process. The coloured-stone trade has plenty of the first and none of the second.

This is not a criticism of Sotheby's — auction houses do exactly what they are built to do, brilliantly. It is an observation about infrastructure. When equities move 45% on earnings, you can decompose the move by the minute. When a ruby necklace moves 45% over estimate, the trade updates its priors by anecdote: dealers will quote this sale in negotiations for the next six months, insurers will nudge schedules, and nobody will be able to say what a comparable stone is worth today, because between auction seasons there is no today.

The number that's missing

What would it take for a headline sale to become usable market data? The same three things every other asset class already has: continuous quotes between events (so a hammer price lands against a visible baseline, not a void); public comparables (per-stone records of certified weight, grade and origin, so "a ruby sold well" becomes "untreated Burmese material at X per carat"); and an audit trail (valuations that update on the record, not in private schedules).

That is the machinery we're building at CaratShares — per-stone on-chain order books with appraised NAV recorded alongside, aggregated into the Coloured Gemstone Index, a daily per-carat series across 18 species (testnet demo today, methodology open). Against a baseline like that, a September 17 would stop being a mystery and start being a data point.

The trade will keep celebrating great hammers — as it should. The infrastructure question is what happens on all the days in between.

Sources: Sotheby's results; Rapaport, "Sotheby's Scores $34M at Hong Kong Sale," September 2026.

The Coloured Gemstone Index · What a $30M ruby tells us about price discovery · Why coloured gemstones never had a public market